102

How can I

compare (without calculator or similar device) the values of $\pi^e$ and $e^\pi$ ?

darij grinberg
  • 16,217
  • 4
  • 41
  • 85
Mirzodaler
  • 1,277
  • 2
  • 9
  • 8
  • 1
    More generally: http://math.stackexchange.com/questions/517555/fastest-way-to-check-if-xy-yx – Henry Jan 30 '16 at 14:07

15 Answers15

198

Another proof uses the fact that $\displaystyle \pi \ne e$ and that $e^x > 1 + x$ for $x \ne 0$.

We have $$e^{\pi/e -1} > \pi/e,$$

and so

$$e^{\pi/e} > \pi.$$

Thus,

$$e^{\pi} > \pi^e.$$

Note: This proof is not specific to $\pi$.

David
  • 113
  • 15
Aryabhata
  • 79,315
  • 8
  • 179
  • 262
  • 25
    This is from [The Book](http://en.wikipedia.org/wiki/Paul_Erdős#Personality)! Impeccable proof. – Prism Sep 14 '13 at 22:20
  • 1
    if x is negative than would your first equation which is e^x > 1 + x be true ? – Murtuza Vadharia Jan 15 '15 at 17:36
  • 3
    @MurtuzaVadharia: Yes, it is true. Consider $f(x) = e^x -1 -x$. It's derivative is $e^x -1$ which is $\lt 0$ for $x \lt 0$ and $\gt 0$ for $x \gt 0$, so $f$ decreases from $-\infty$ to $0$, and increases from $0$ to $\infty$. Since $f(0) = 0$... – Aryabhata Jan 21 '15 at 08:11
  • 6
    I think it's worth mentioning that this works for all positive numbers which aren't $e$. The proof has nothing to do with $\pi$. – Nikolaj-K Feb 07 '15 at 20:48
  • @NikolajK: Thank you for the suggestion! Done. – Aryabhata Feb 10 '15 at 00:52
  • Can you please explain how did you get $e^{π/e−1}>π/e$? – MrAP May 28 '17 at 08:49
  • @MrAP: $ x = \frac{\pi}{e} -1$. Now use $e^x \gt x + 1$. – Aryabhata May 30 '17 at 20:20
  • How did you get the intuition to make the substitution $x = \frac{\pi}{e} - 1$? I understand how to get to the step $e^x > 1 + x$, but I wouldn't have thought of the substitution for $x$ so quickly. Is there some kind of obvious procedure you followed? – David Jul 21 '20 at 18:47
  • 1
    @David: Sorry to be of no help :(. No idea what the thought process was, it has been quite a while. – Aryabhata Aug 03 '20 at 20:06
51

This is an old chestnut. As a hint, it's easier to consider the more general problem: for which positive $x$ is $e^x>x^e$?

Robin Chapman
  • 21,512
  • 2
  • 59
  • 78
46

From Proofs without Words.

alt text

Andrey Rekalo
  • 7,524
  • 3
  • 45
  • 43
43

Alternatively, we can compare $e^{1/e}$ and $\pi^{1/\pi}$.

Let $f(x) = x^{1/x}$. Then $f'(x) = x^{1/x} (1 - \log(x))/x^2$. Since $\log(x) > 1$ for $x > e$, we see that $f'(x) < 0$ for $e < x < \pi$. We conclude that $\pi^{1/\pi} < e^{1/e}$, and so $\pi^e < e^\pi$.

The same calculation shows that $f(x)$ reaches its maximum at $e^{1/e}$, and so in general $x^e < e^x$.

Yuval Filmus
  • 55,550
  • 5
  • 87
  • 159
19

Let $f (x) =$ $x^\frac1x$

Find value of $x$ such that function gets maximum value

For this functions for $x=e$ function will get the maximum value

so $e^\frac1e$ is greater than $\pi^\frac1\pi$

so $e^\pi$ is greater than $\pi ^e$.

Murtuza Vadharia
  • 1,550
  • 4
  • 15
  • 31
16

Elaborating Robin's answer take $f(x) = \log{x} - \frac{x}{e}$. We have $$f'(x)= \frac{e-x}{xe}$$ Thus $f'(x)>0$ for $0 < x < e$ and $f'(x) <0$ if $x > e$. Consequently, we have $f(x) < f(e)$ if $x \neq e$.

Exercise: Try to prove this using the same methods: $2^{\sqrt{2}} < e$.

11

Hint:

Prove that the function $f(x)=\frac{e^x}{x^e}, x\geq e$ is strictly increasing on the interval $x\in \left [ e,\pi \right ]$. What is $f(e)$ and $f(\pi)$?

EricAm
  • 990
  • 2
  • 14
  • 27
9

Another visual proof. Recently published in arXiventer image description here

nbsrujan
  • 211
  • 2
  • 7
3

Yet another line of thought would be this: $$\begin{align} e^{\pi} > \pi^e &\iff \pi \ln(e)>e\ln(\pi)\\ &\iff \pi >e\ln(\pi) \\ &\iff \ln(\pi)>\ln(e)+\ln (\ln (\pi)) \\ &\iff \ln(\pi)>1+\ln (\ln (\pi)) \end{align}$$ By concavity of $\ln$, $x-1>\ln(x)$ for all $x\neq 1$. With $x=\ln(\pi)$ we get the inequality wanted.

Gabriel Romon
  • 32,348
  • 5
  • 56
  • 135
2

\begin{align} &e^\pi>\pi^e \\[5pt] \iff&\exp(\pi)>\exp(e\log\pi) \\[5pt] \iff&\pi>e\log\pi \\[5pt] \iff&\frac{\pi}{\log\pi}>e \\[5pt] \iff&\frac{\pi}{\log\pi}>\frac{e}{\log e} \end{align} The final line is true because the function $\dfrac{x}{\log x}$ is strictly increasing on $[e,\infty) \, ,$ and the result follows.

Joe
  • 14,185
  • 2
  • 28
  • 65
1

Denote $n = e^\pi$, $m = \pi^e$ and $s = \log \pi$. Then $\log n = \pi = e^s$ and $\log m = e \log \pi = es$. Then $$\log \frac {n} {m} = \log n - \log m = e (e^{s - 1} - s).$$ By Taylor expansion, we have $$e^{s - 1} = 1 + (s - 1) + \cdots > s.$$ Then $$\log \frac {n} {m} = e (e^{s - 1} - s) > 0.$$ Hence, $n > m$.

1

Just adding a recent one. Another visual proof. I am doubtful if it has open access.

https://link.springer.com/article/10.1007/s00283-019-09964-x

1.414212
  • 165
  • 10
1

Not that this question needs another answer, but here is a proof of $e^\pi > \pi^e$ using the Mean Value Theorem applied to $\ln x$ on the interval $(e,\pi)$, along with the assumptions that $e<\pi$, $\ln$ is increasing, and $\frac{d}{dx} \ln x = \frac{1}{x}$.

By the MVT, there exists $c \in (e,\pi)$ such that $$ \frac{\ln \pi - \ln e}{\pi - e} = \frac{1}{c} $$ We can increase the right hand side by replacing $c$ with the small number $e$, and so we have $$ \frac{\ln \pi - \ln e}{\pi - e} < \frac{1}{e} $$ and thus $$ {\ln \pi - \ln e} < \frac{1}{e}(\pi-e) $$ or $$ {\ln \pi - 1} < \frac{\pi}{e}-1 $$ which gives $$ e \ln \pi < \pi \ln e$$ and therefore $$ e^\pi < \pi^e$$

I. J. Kennedy
  • 3,790
  • 3
  • 24
  • 39
0

Let

$$f(x) = e^x$$

$$G(x) = x^e$$

We can simply show that

$$f(e)=G(e)$$

$$f'(e)=G'(e)$$

For $x > e$ the $f(x)$ will grow faster than $G(x)$

Then

$$e^{\pi} > \pi^{e}$$

Pentapolis
  • 561
  • 3
  • 9
-6

$$e^\pi>\pi^e$$ because if we subtract $e$ from both exponents we get $$e^{\pi-e}>1$$ which is true because $e$ is is greater than $1$ so when it 8s raised to that power it is equal to $$\frac{e^\pi}{e^e}$$ and that has to be greater than 1 because the top is greater than the bottom.

Ѕааԁ
  • 31,920
  • 22
  • 33
  • 72
user533594
  • 11
  • 2
  • Are you sure you don't want to reconsider that first assertion? $\pi^e>e^e$. –  Mar 03 '18 at 05:44
  • 4
    You can't subtract $e$ from both exponents unless the base is the same. – Zaz Apr 14 '19 at 01:13